Efficient Coding Provides A Better Account Of Systematic Biases In Locomotor Space Perception Than Does Action Ability

Document Type

Poster Session

Publication Date


Published In

Journal Of Vision


Quantitative predictions of egocentric distance and geographical slant perception may be derived from dense coding (scale expansion) in angular variables (Durgin & Li, 2011). In particular, distance/height matching tasks can be accounted for by measured biases in perceived direction (elevation) relative to the horizon (Li et al., 2011). However, substantial interest has accumulated for studies arguing that perception is affected by physiological factors such as BMI and age. We used a cross-sectional methodology to look at space perception across the adult lifespan (ages 18-72). We examined performance in several distance, height, and slant tasks including estimation and perceptual matching. We also measured factors concerning specialized knowledge (golfers and skiers know more about distance and slope estimation, respectively), physiology (age, BMI), and personality (e.g., Agreeableness) that might influence performance. Participants (N=106) included college students (N=58), but also non-student members of the nearby community (N=48). Consistent with coding efficiency theory, slant, distance and height estimates closely accord with the angular expansion account: Distances were underestimated (0.72 of actual distance); height was overestimated (1.11 of actual height) and egocentric distance-height matching performance (mean ratio: 1.53) accorded well with the ratio between estimate ratios (i.e., 1.54). Contrary to the action-ability hypothesis, participants defined as obese by CDC guidelines (mean BMI = 35.2) tended to underestimate distance more than those defined as normal. Slant overestimates were consistent with scale expansion, but were also affected reliably both by knowledge and by age. Although obese adults tended to give higher slant estimates, the reliable effect of age on slant estimation was opposite to the predictions of action ability (older adults gave lower slant estimates). Whereas some aspects of the slant results could be used to support an action-ability account, the data as a whole contradict the action-ability account. The data quantitatively support the coding efficiency account.


Vision Sciences Society Fifteenth Annual Meeting

Conference Dates

May 15-20, 2015

Conference Location

St. Pete Beach, Florida